A new metastable convergence criterion and an application in the theory of uniformly convex Banach spaces (Q2320042)

This paper is concerned with proof mining, a research paradigm that aims to analyze proofs in mainstream mathematics using proof-theoretic tools in order to gain additional (usually quantitative) information. The author analyzes a strong convergence proof of \textit{W. A. Kirk} and \textit{B. Sims} [Bull. Pol. Acad. Sci., Math. 47, No. 2, 147--155 (1999; Zbl 0977.47044)] for the Picard iteration for asymptotically nonexpansive mappings which generalizes a `nonempty interior' argument from Hilbert spaces to uniformly convex Banach spaces. As it is usually the case in this sort of endeavours, a~uniform and computable rate of convergence for the iterative sequence \((x_n)\) may not exist, so one then turns to the next best thing which is the concept of a rate of metastability, i.e., a bound on the \(N\) in the statement \[ \forall \varepsilon > 0 \; \forall g: \mathbb{N} \to \mathbb{N} \; \exists N \; \forall i,j\in [N,N+g(N)] : \|x_i - x_j\| \leq \varepsilon, \] depending on the \(\varepsilon\) and the \(g\), and also on some parameters specific to the problem at hand. Kirk and Sims [loc.\,cit.]\ observed that the convergence argument only depends on a `convergence to the infimum' condition. The author further simplifies this into an equivalent form which he dubs `asymptotic decreasingness'. He then defines a corresponding metastable version of asymptotic decreasingness with an associated rate and shows that such a rate can yield (in conjunction with a modulus of uniform convexity) a rate of metastability for the iteration under study.